The threat of technological unemployment

There is a widely reported threat to our economy: robots are going to replace human workers. It is nothing new… In 1930, Keynes, the famous economist introduced the term “technological unemployment”…

We are being afflicted with a new disease of which some readers may not yet have heard the name, but of which they will hear a great deal in the years to come—namely, technological unemployment. This means unemployment due to our discovery of means of economising the use of labour outrunning the pace at which we can find new uses for labour.

There is a long Wikipedia article on technological unemployment, where you can learn that people have been concerned about it for centuries… long before we could even imagine robots and computers. It is a recurrent theme of western thought.

Personally, I am unconcerned by technological unemployment as a threat. On the long run, I think that we can create jobs out of thin air if we need to (something I call “transemployment”). I strongly disagree with people like Tyler Cowen who think that regular folks are doomed.

We should first recognize the thesis for what it is. Currently, most people get most of their income through jobs. But jobs provide a lot more than just income. They also provide meaning and a social status. In some countries, like the USA, people often get access to critical health services through their employment. Unemployed people are more likely to be sick or depressed, they are less likely to be seen as viable mates… and so forth. We live in a society where even billionaires have “jobs”. If human labor were to become wholly unnecessary, we are going to substitute for it through transemployment. That is, we will create work that is not strictly needed. I think that this process is well under way. But how will we pay for it all? Well. Consider that if human labor becomes unnecessary, it follows that the wealth is being created by the robots, so it is still there. Some authors fear that technology will result in a radical concentration of wealth, the like of which we have never seen… while a few people will be super wealthy, all of us will slowly starve. Except that fewer people than ever in history are starving! It is true that people without a high school education in the US earn much less than the more educated. And there is an income gap building up. However, in the US, the less educated also enjoy a much higher amount of leisure time. Some of us envy the CEOs, but let us not forget that many of them work 60 hours or more. Do we really want all to focus our lives almost entirely on “jobs” forever and ever? Can’t we imagine a world where jobs are relatively secondary aspects of our lives? Maybe in the future, when we speak with teenagers, we will never mention employment as something they need to be concerned with.

I think that we can’t automate fast enough. I work for a university where too many people spend too much time on routine work, filling out the same forms day after day. We can’t seem to outrun bureaucracy with automation. Hospitals can’t find enough qualified nurses willing to work at modest wages: we need to more automation in health care, and we need it now. China is running out of young factory workers, they can’t get the robots in place soon enough.

I really do wish that we were much further along regarding technological unemployment. My bias as a techno-optimist is to be bullish on technology… but let us look at the signs… is technological unemployment something that we can observe around us?

  • Though the economy goes up and down, the unemployment rate worldwide is low. In is under 5% in the US, and even lower in Japan. I don’t think that people advocating imminent technological unemployment have a good explanation for this data point. Yes, some people get discouraged and drop out of the job market, thus artificially lowering the unemployment rate, but that has always been true.
  • In the USA, the participation rate, that is the fraction of people who work in the entire population is lower than it was 10 years ago by about 3%. In the last 10 years, the participation rate has been flat in France and in Japan, declining in the US and Canada, while it has been rising in Germany. The participation rate varies greatly from country to country. It is 56% in France, 60.5% in Japan and Germany, 63% in the US and 66% in Canada. It seems difficult to reason about it in absolute terms.

    In the US, a lot of people in their “prime age” who are not working are sick in some way….

    Participation in the labor force has been declining for prime age men for decades, and about half of prime-age men who are not in the labor force (NLF) may have a serious health condition that is a barrier to work. Nearly half of prime age NLF men take pain medication on a daily basis, and in nearly two-thirds of cases, they take prescription pain medication. The labor force participation rate has stopped rising for cohorts of women born after 1960. Over the past decade, retirements have increased by about the same amount as aggregate labor force participation has declined. Continued population aging is expected to reduce the labor force participation rate by 0.2 to 0.3 percentage point per year over the next decade. (Krueger, Where Have All the Workers Gone? 2016)

    But sickness and disability affect disproportionally older people.

    Young people outside of the workforce are sometimes “discouraged”, meaning that they do not seek employment because they see no hope of getting a good job. The number of discouraged people in the US has been slowly going down since its latest peak in 2010.

  • Firm investments in high-tech equipment and software is falling, relatively speaking. That’s hardly what you’d expect in an economy where robots are taking over? Of course, this is partially explained by the rapidly falling cost of some technologies.
  • Worker productivity has been increasing more or less monotonically for the last few decades but it has recently been declining. A technological boost akin to what might lead to technological unemployment should be accompanied by a rise in worker productivity.
  • When I was born, everyone dealt with a living and breathing bank teller. These days, I go see a bank teller maybe once a decade. They have been “replaced” by machines. But did you know that there are more bank tellers in the US than ever before? And I’d argue that bank tellers today have more interesting work, doing a lot less in terms of routine work.

Overall, there is scant evidence that we are undergoing a technological-unemployment crisis, if only because unemployment rates are low.

Daniel Lemire, "The threat of technological unemployment," in Daniel Lemire's blog, December 26, 2016.

Published by

Daniel Lemire

A computer science professor at the University of Quebec (TELUQ).

18 thoughts on “The threat of technological unemployment”

  1. I would like to see an update to this post one year after the first driverless truck crosses the roads in a regular trip.

    With update I mean a postdata or a new post.

    With regular trip I mean that there’s no human driver supervision, it’s not testing, etc

  2. “we will create work that is not strictly needed” – maybe in Canada, but good luck with that in the USA, especially after another tuck towards laissez-faire capitalism with Trump presidency.

    “in the US, the less educated also enjoy a much higher amount of leisure time” – especially these with two or three low paying jobs.

    My point is that countries like Canada, with well developed social safety nets have a chance dealing with advanced automation. Countries like the USA have a high chance of turning it into nightmare. This is not a technical or economics problem, but the one of social structure.

    1. “in the US, the less educated also enjoy a much higher amount of leisure time” – especially these with two or three low paying jobs.

      You might think that people who are on the lower scale of income work longer hours, but the reverse is true. Statistically, the people doing the equivalent of two jobs are in the top quintile of income:

      The annual number of hours of employed labor in the top quintile is still nearly twice that in the bottom quintile. (Hederman & Rector)

  3. Employment is a side-effect of capitalism’s need to generate saleable products. Assuming there is market demand for your products, which the ‘millennial’ generation has shown many companies the door, when you don’t have a need for labor to produce goods, you invite oligopolies to flourish. The fastest, biggest, richest will be able to hit economies of scale first and push or buy others out of the market, until you have no market left. Technology exacerbates this problem. Labor and spending money are the individuals’ only sources of power in the economy. When labor has less and less impact on a company’s bottom line, labor gets treated badly more and more. Without government incentive to create jobs and without the philosophical idealism surrounding work, there will be nothing to stop companies from automating as much as they can. It will be up to governments and communities to decide how they will let companies treat people and the environment when they the people no longer have any power in their economic system.

  4. There’s a lot of talk about technological unemployment, and I think we’re possibly in the ‘uncanny valley’ stage. You can see where things are headed, but it’s not good enough to actually use. Yet.

    But the issue is really sociological, just as it was with the Luddites and the looms – the technology itself wasn’t the problem, but the social structure of who owned the looms, etc.

    Countries that are structured for the needs of people will be able deal with automation, those that are modelled on feral capitalism, less so.

  5. I’d like to raise an issue with your bank teller point. Your reference uses a reference which indicates that bank teller employment has increased substantially since the 70s (when ATMs were just introduced), but since the 80s (when, I assume, they began to be used in earnest) only by about 16% (from 500k to say 580k). On the other hand the US population has increased from ~220 million to ~320 million (and the labor force has had a correspondingly large increase). Although true, isn’t it a little misleading to say that there are more tellers than ever? It seems like it would be a harder job to get with the change in ratios, for instance.

    1. @Roderic

      I don’t think that the statement is misleading. On the contrary, I think it is very much a “leading” observation.

      Consider truck drivers. We now have self-driving trucks. Some people say that it implies that soon these drivers might be out of a job. Are they correct?

      What if, in 30 years, we had more truck drivers than today? What the bank teller story tells you is that this is entirely possible, even with fully automated trucks.

      Why might this be? Well, bank tellers in 2016 help people with non-mundane tasks. Most of the simple cash withdrawals are automated… but there are many things that are easier and simpler with a human being simply because it exceeds the programming of the machine.

      What might happen is that number of deliveries might multiply… say become 10x what it is today, so that truck drivers would occupy a small niche, but this small niche could be larger than the whole truck delivery industry today.

      The trick is that automation works best by covering the easiest 80% of all cases. It gets increasingly more difficult to automate whatever remains. At a glance, it seems like bad luck for the human beings because they get corned to the remaining 20%… but the whole industry might grow so large that this 20% is larger than the whole industry when automation started out.

  6. Thank you for the lengthy, considered reply!

    I appreciate that the insight you describe was what you were leading to with the bank teller point, and now the truck driving hypothetical, however I don’t think having “more” of a job is necessarily a sign of harmlessness for automation. Could we have 1 more bank teller job than 1980 and 50% more workers and make the same conclusion? More jobs doesn’t mean effectively more work for the people who fill those jobs if proportionally less of them are needed.

    Bank teller jobs grew, but not nearly at pace with the growth of the whole labor force. If truck driving and other sorts of jobs are partially automated and grow at a similar pace, wouldn’t we expect the sort of people who fill those jobs now to be much worse off in the future with the higher competition (assuming population / labor force growth is still much greater)?

    I also appreciate that automation may create entirely new jobs. Hopefully industry growth would include adding many of these new jobs and fill in the missing job growth, but your reference doesn’t indicate whether that was the case or not for banking. My point is to take issue with using data that only looks at one job and indicates hugely slowed growth for that job after automation is introduced and treating it as a positive.

  7. It seems quite plausible to me that society can find a stable organization with massive automation of routine/menial jobs. However I’m concerned about the pace. It took in the (very rough) ballpark of a century for developed economies to go from the majority of the population working in agriculture to only a few percent. That’s several generations! If the more aggressive predictions about the AI-related changes we are on the cusp of are even close to correct, we’ll see similar economic migrations in a single generation.

    People and their institutions don’t have a great track record at adapting to such big changes that quickly.

  8. I just don’t believe there will be jobs for humans soon. We have a finite skill set. We have vision, hands to manipulate objects, legs to get us around, the ability to communicate and understand language, and a brain to problem solve. Machines are making rapid progress in all of these areas. Once you have an intelligent machine that can move around and manipulate objects…why would you then ever hire a human? Some say we will have that kind of intelligence by 2035. People believe new industries will be created through technology as this has always happened in the past so people will work in those new fields. However, the past never had super intelligent AI and machines with human like bodies. Once we have these things why would you ever hire a human who is lazy, incompetent, physically and intellectually weaker?

  9. Its true that tech has created more jobs than it destroyed in the past. But i think tech vis a vis employlent moves in three stages. First stage: It will create more jobs than it destroys. Second stage: As it gets more advanced it will slowly begin to destroy as many jobs as it will create. Third stage: As it gets even more advanced it will destroy more jobs than it will create.

    We have to figure which stage the tech is in.

  10. I have read stories of how automation is penetrating the indian software industry. A google search about this will tell you the whole story.

      1. Not a decline, but i remember reading that fewer and fewer software programmers will be needed to produce the same amount of revenue for the companies. For eg: 100 billion revenues by software companies were achieved using five million employees upto the year 2015. The companies’ top management predict that the next billion in revenue can be achieved by only 3 million using software automation and other kinds of tech! Thats a potential job loss of two million in the software industry. This also have more wider effects. For example india’s domestic construction and retail sectors depend a lot on the new rising middle class which are made up mostly by these software engineers. So these two million potential job losses will affect jobs in these two sectors too. And in many more sectors like clothing, airlines, electronics….you name it. I dont remember these as the exact numbers but the percentages are around the same ballpark. You could google it though (sorry :)).

        Another example: Since the rise of online shopping in india clothing has become very cheap. Due to this many more are buying more clothing than before. This is leading?lead? to more jobs initially. But another dangerous trend has begun. The price depressing effects of these companies has led to a fall in clothes prices in the retail sector. Now what will happen because of this is, this will lead to less hiring in retail. It will also lead to more pressure on cloth suppliers and textile factories to reduce their prices( who will do it by reduce their staff) and by putting more pressure on cotton traders to lower their prices who will put pressure on cotton farmers to lower their prices.

        All these sectors are beginning to get affected slowly.

        1. Not a decline, but i remember reading that fewer and fewer software programmers will be needed to produce the same amount of revenue for the companies.

          We are supposed to get more productive over time. If this did not happen, most of us would still work on farms, trying to keep our kids fed.

          As workers can do more things in less time, it allows us to pursue new services and products.

          In 1990, hardly anyone in India worked on software or the Internet. Today, thousands do… but that was made possible because these same people are not needed to produce food and housing.

          In time, we will need fewer people to keep our servers running… and this means that these people will be free to do other work.

          It is a good thing.

          Another example: Since the rise of online shopping in india clothing has become very cheap. Due to this many more are buying more clothing than before. This is leading?lead? to more jobs initially. But another dangerous trend has begun. The price depressing effects of these companies has led to a fall in clothes prices in the retail sector. Now what will happen because of this is, this will lead to less hiring in retail. It will also lead to more pressure on cloth suppliers and textile factories to reduce their prices( who will do it by reduce their staff) and by putting more pressure on cotton traders to lower their prices who will put pressure on cotton farmers to lower their prices.

          You are pretty much describing how capitalist economies get richer over time. Things get cheap, people get more productive and so forth.

  11. I really like this article, it offers me some new perspectives. But I’m wondering how could some developing countries whose people rely on low-paid jobs cope with this issue since their people are very likely to lose their jobs. Maybe, in the long run, technological development can help better earth residents as a whole, but in the short run, it’s a bit worrying. anyway, I’m also optimistic as well. But I think there’s a lot governments, international communities have to do to help make this inevitable trend something positive. hope to read more about this issue.

  12. The threat of technological unemployment or, at a minimum, severe underemployment is very real in the United States and other countries with strong negative views of proactive social safety nets.

    The McKinsey Global Institute estimates that 50% of current work activity will automatable, replacing as many as 800 million jobs globally by 2030. The White House Council of Economic Advisors finds that more than 80% of workers earning less than $20 per hour will lose their jobs to automation by 2036. 48% of 1,896 AI experts surveyed by Pew Research believe that AI and robotic technologies will displace more jobs than they create by 2025.

    There is a strong likelihood that society will “create work that is not strictly needed,” but what will this work pay? When work fails to create economic value, by its very nature it will be less financially rewarded. This will, at a minimum, create an extreme economic disparity between the techno-capitalists that employ the robots, and the people that do the meaningless work simply to have something to do.

    Forward-thinking countries need to immediately explore universal basic income or negative income tax to counter technological unemployment. The number of humans on this planet continues to grow, and there will undoubtedly be less meaningful work for humans in the future as AI and robotics replace the need for human labor.

Leave a Reply

Your email address will not be published.

You may subscribe to this blog by email.